When propaganda encounters backlash from the public, how does the authoritarian regime adapt its propaganda strategies?

While a substantial body of research has highlighted the effectiveness of authoritarian pro- paganda in shaping the information landscape and swaying public opinion (Huang, 2015; Weiss and Dafoe, 2019; Carter and Carter, 2021; Mattingly and Yao, 2022; Pan et al., 2022; Rozenas and Stukal, 2019) and limitations (Huang, 2018; Liao and Hwang, 2022; Potter and Wang, 2022), how the authoritarian propaganda adapts to backlashes have been understudied.

We collected propaganda data from street-level bureaucracy WeChat accounts, correlating over 90000 posts with over 5000 citizen petitions from February to May, both by date and street.

• We used Structural Topic Modeling (STM) and BERTopic to detect if the petitions and propaganda posts are related.

• We used the similarity score based on the Sentence Transformer to capture the dynamics of propaganda.